A Maturity Model for Enterprise Data Quality ... - Semantic Scholar

5 downloads 92884 Views 870KB Size Report
2008). Typical classes of master data are supplier ..... The Maturity Model was built in the course of ..... A German supplier from the automotive in- ..... Technics Publications, Denver,. Colorado. DAMA (2009) The DAMA Guide to the Data Man-.
Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 Martin Ofner, Boris Otto, Hubert Österle

4

Martin Ofner, Boris Otto, Hubert Österle

A Maturity Model for Enterprise Data Quality Management Enterprises need high-quality data in order to meet a number of strategic business requirements. Permanent maintenance and sustainable improvement of data quality can be achieved by an enterprise-wide approach only. The paper presents a Maturity Model for Enterprise Data Quality Management (Enterprise DQM), which aims at supporting enterprises in their effort to deliberately design and establish organisation-wide data quality management. The model design process, which covered a period of five years, included several iterations of multiple design and evaluation cycles and intensive collaboration with practitioners. The Maturity Model is a hierarchical model comprising, on its most detailed level, 30 practices and 56 measures that can be used as concrete assessment elements during an appraisal. Besides being used for determining the level of maturity of Enterprise DQM in organisations, the results of the paper contribute to the ongoing discussion in the information systems (IS) community about maturity model design in general.

1

Introduction

Data quality management (DQM) as an organisational function comprises all practices, methods, and systems for analyzing, improving and maintaining the quality of data. DQM basically aims at maximizing the value of data (customer data, supplier data, or material data, for example) (DAMA 2008). Over the last 15 years DQM has been the subject of analysis in many publications both by researchers (Batini and Scannapieco 2006; Otto et al. 2007; Wang 1998; Wang et al. 1998) and practitioners (English 1999; Loshin 2001; Redman 2000). Although data quality is widely recognized as a strategic success factor, the majority of companies consider DQM in their organisation as ‘being in the early phases of maturity’ (Pierce et al. 2008). Particularly certain business requirements, such as effective supply chain management (Kagermann et al. 2010; Tellkamp et al. 2004; Vermeer 2000), improved decision-making (Price and Shanks 2005; Shankaranarayan et al. 2003), compliance with legal or regulatory provisions (Friedman 2006; Salchegger and Dewor 2008), or efficient customer relationship management (Reid and Catterall 2005; Zahay and Griffin 2003)

demand an enterprise-wide approach to DQM, as such requirements cannot be met by isolated solutions or single business units alone. In order to be able to establish enterprise-wide DQM in the following referred to as Enterprise DQM , changes are needed on a strategic, on an organisational, and on an information systems level (Baskarada et al. 2006; Bitterer 2007; Lee et al. 2002; Ryu et al. 2006). In their effort to bring about these changes companies need support and assistance, particularly with regard to monitoring the progress in establishing Enterprise DQM. Taking this into account, the research question examined in this paper is how companies may deliberately design Enterprise DQM. The word deliberately refers to the need that companies are capable of identifying areas for improvement and deriving appropriate action with regard to Enterprise DQM. The research objective is to design a model that allows assessing the maturity of Enterprise DQM, with the research process following the principles of design science research (Hevner et al. 2004; Österle and Otto 2010). Maturity models support organisational change insofar as they represent an instrument for decision-

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 A Maturity Model for Enterprise Data Quality Management

makers to assess an organisation‘s actual state, derive actions for improvement, and evaluate these actions afterwards in terms of their effectiveness and efficiency (Crosby 1979; Gibson and Nolan 1974; Nolan 1973). The following section of the paper outlines the theoretical foundations underlying the research and compares existing maturity models from the DQM domain. After that the research methodology and the process of designing the Maturity Model for Enterprise DQM are elaborated. Then the design rationale of the structural specification of the Maturity Model (i.e. the conceptual model) is discussed, alongside with procedural guidelines for applying this conceptual model. Afterwards, a first evaluation of the Maturity Model is provided, and findings and implications are discussed. The paper concludes with a short summary and recommendations for further research on the topic.

2 2.1

Theoretical Foundations Data and Data Quality

Singular pieces of data specify discrete characteristics of objects and processes from the real world. In this sense, data is free of context (Boisot and Canals 2004; Davenport and Prusak 1998; Spiegler 2000). Business distinguishes between master data and transaction data. Master data consists of attributes describing a company’s core business objects. It constitutes the basis for both operative value creation processes and analytical decision-making processes (Smith and McKeen 2008). Typical classes of master data are supplier master data, customer master data, or product master data (Mertens 2000). Transaction data describes business processes. It relates to master data, and therefore its existence is dependent on this master data (Dreibelbis et al. 2008). It is master data that is of particular importance to Enterprise DQM, as the quality of such data is critical for meeting the business requirements mentioned above. Thus, master data needs to be defined for the whole of an organisation and must allow to be identified unambiguously.

5

When data is used within a certain context or when data is processed, it turns into information (Boisot and Canals, 2004; van den Hoven, 2003). Although the terms data and information are clearly distinguished in theory, a clear definition on what quality means to either aspect does not exist. Both information quality and data quality is seen as a context dependent, multi-dimensional concept, describing the ‘fitness for use’ of information and data as determined by a user or user group (Wang 1998). The fact that information quality and data quality is considered to be context dependent emphasizes the notion that it is up to the user to decide whether certain information or data is useful (Wang and Strong 1996). Hence, ‘fitness for use’ can be perceived in different ways, manifesting itself in so-called data quality dimensions. Numerous scientific studies have dealt with the identification and description of such data quality dimensions (Price and Shanks 2005; Wand and Wang 1996; Wang and Strong 1996; Wang et al. 1995). Among the most important ones are accessibility, accuracy, completeness, and consistency (DAMA 2008).

2.2

Data Quality Management

Data Management Association (DAMA) defines DQM as ‘application of Total Quality Management (TQM) concepts and practices to improve data and information quality, including setting data quality policies and guidelines, data quality measurement (including data quality auditing and certification), data quality analysis, data cleansing and correction, data quality process improvement, and data quality education’ (DAMA 2008). DQM aims to achieve the following goals: establish DQM as an organisational function, design DQM to cover the organisation as a whole, establish a continuous improvement process for DQM, qualify and authorize staff for executing DQM tasks, provide appropriate techniques and guidelines for DQM (Batini and Scannapieco 2006; English 1999; Wang 1998; Zhang 2000). In order to emphasize the imperative to establish DQM in an enterprisewide approach, the paper at hand refers to DQM as Enterprise DQM.

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 Martin Ofner, Boris Otto, Hubert Österle

6

2.3

Maturity Models and Organisational Change

Maturity models represent a special class of models, dealing exclusively with organisational and information systems related change and development processes (Becker et al. 2010; Crosby 1979; Gibson and Nolan 1974; Mettler 2010; Nolan 1973). Maturity models consist of an organized set of constructs serving to describe certain aspects of maturity of a design domain (Fraser et al. 2002). The concept of maturity is often understood according to the definition of Paulk et al. (1993), who consider maturity to be the ‘extent to which a process is explicitly defined, managed, measured, controlled, and effective’. Most maturity models explicitly or implicitly follow this definition, taking a process oriented view when looking at how a design domain can be assessed and optimized. The sole focus on the process perspective has been controversially discussed in literature (Bach 1994; Gillies and Howard 2003; Jones 1995; Pfeffer and Sutton 1999). What is demanded by critics of this approach is an all-encompassing, integrated concept for measuring levels of maturity, taking into account technological and cultural aspects as well (Christensen and Overdorf 2000; Saleh and Alshawi 2005). Typically, a maturity model consists of a domain model and an assessment model. The domain model comprises criteria by which the design domain can be partitioned into discrete units to be assessed. The assessment model provides one or multiple assessment dimensions, each of which defining an assessment scale. What is basically assessed is to which extent certain criteria comply with the scale for each assessment dimension. In order to structure the assessment process some maturity models also provide appraisal methods (e.g. Standard CMMI Appraisal Method for Process Improvement, SCAMPI) (SEI 2006b). Basically, two types of maturity models can be distinguished. Staged models build on best practices to explicitly specify an ideal path of development of a design domain (Paulk et al. 1993). Continuous models are used to review

certain quality features of a design domain at regular intervals, determine the level of maturity for different features or criteria, and derive actions for improvement. In the case of continuous models the path of development is dynamic, i.e. it is not predefined by the model (EFQM 2009).

3 3.1

Related Work DQM Approaches

In recent years a number of methods have been developed both by the research and the practitioners’ community supposed to offer support and assistance in selecting, adapting and applying techniques for improving data quality (Batini et al. 2009). These methods describe best practices for the DQM domain and can be used to derive criteria for designing a Maturity Model for Enterprise DQM. The Complete Data Quality Methodology (CDQM) sees DQM as being composed of a series of singular projects for data quality improvement (Batini and Scannapieco 2006). These projects are results oriented, i.e. the data quality to be achieved is put in relation to the costs that are likely to occur in the process. Only those projects are realized which promise to be reasonable and profitable from a business perspective. Redman (2000) developed the Data Quality System (DQS), focusing on the provision of an organisational framework (strategy, training concepts, etc.) and the development of business and technical capabilities (data quality planning, data quality measurement, data models, etc.). Total Data Quality Management (TDQM) is the name of a research program at the MIT. TDQM sees information as a product (known as the information product (IP) approach) that needs to be produced according to the same principles physical goods are produced, including exact specification of requirements to be met by information products, control of the production process along the entire lifecycle of information products, and naming of an information product manager (Wang 1998; Wang and Strong 1996).

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 A Maturity Model for Enterprise Data Quality Management

Total Quality data Management (TQdM) is a method that offers support when information needs to be optimized for business purposes (English 1999). TQdM follows the principles of the IP approach and focuses even more on the definition of requirements to be met by information products. To sum up, it can be said that all of these methods refer to results oriented, cultural, process related, or technological aspects of data quality management.

3.2

Maturity Models for DQM

Beside the methods described in the previous section also maturity models for DQM have been developed. Lee et al. (2002) have proposed a methodology for information quality assessment (AIMQ), which can be used as a basis for information quality assessment and benchmarking. This methodology uses 65 criteria to evaluate results to be achieved by DQM. DataFlux (2007) has come up with a maturity model comprising four criteria (people, policies, technology, and risk & reward) by which companies can assess the progress of DQM establishment in their organisation. Bitterer (2007) aims at the same objective with their maturity model, using quite vague definitions of individual levels of maturity instead of clearly defined criteria. Ryu et al. (2006) and Baskarada et al. (Baskarada et al. 2006) have developed maturity models on the basis of the Capability Maturity Model Integration (CMMI) approach (SEI 2006a). The scope of both models is quite narrow with regard to DQM. While the former defines 16 criteria for specifying and maintaining metadata (which is seen as a prerequisite for achieving high quality of data), the latter focuses on information systems for the mechanical engineering industry, for which it defines 19 technical criteria. As Tab. 1 shows, none of the maturity models examined covers all aspects of Enterprise DQM.

7

Guidelines for designing actions for improvement are offered by two approaches only. Also, all maturity models examined are characterized by a rigid, predefined path of development. This, however, stands in contrast with the view of DAMA (2009) that states ‘[. . . ] how each enterprise implements [DQM] varies widely. Each organisation must determine an implementation approach consistent with its size, goals, resources, and complexity. However, the essential principles of [DQM] remain the same across the spectrum of enterprises [. . . ]’. Taking this into account, a Maturity Model for Enterprise DQM must provide a dynamic path of development, which each organisation may adapt to its individual needs and requirements.

4 4.1

Research Approach Research Method

The work presented in this paper is an outcome of design oriented research, following the methodological paradigm of Design Science Research (DSR). DSR aims at designing artefacts (constructs, models, methods, or instantiations, for example) in order to solve problems occurring in practice (Hevner et al. 2004; March and Smith 1995). The artefact to be constructed is a maturity model that allows to deliberately design Enterprise DQM. When developing a reliable maturity model a critical factor is the level of maturity of the design domain itself. The less developed a design domain is, the higher is the uncertainty in terms of having valid and reliable knowledge about this design domain, and the higher is the need for a maturity model that is capable of guiding the path of development for designing the domain. If this is the case, usually only few cases are available that help identify possible criteria and evaluate the model, resulting in maturity models of limited reliability only. So access to practitioners’ knowledge is critical for being able to define and evaluate relevant criteria. Therefore, the overarching research method selected for designing a Maturity Model for Enterprise DQM is consortium research,

Enterprise Modellingand and Information Systems Architectures Enterprise Modelling Information Systems Architectures

8

No. X, Month 200X Vol. 8,Vol. No. 2,X, December 2013 Martin Ofner, Boris Otto, Hubert Österle Martin Ofner, Boris Otto, Hubert Österle

4

Source

Results oriented criteria

Culture related criteria

Process related criteria

Technologyrelated criteria

Guidelines offered

Path of development

(Lee et al., 2002)

4

0

0

0

No

Staged

(DataFlux, 2007)

0

4

4

4

Yes

Staged

(Bitterer, 2007)

0

2

2

2

Yes

Staged

(Ryu et al., 2006)

0

0

4

0

No

Staged

(Baskarada et al., 2006)

0

0

0

4

No

Staged

Key: 4 = Criteria formally defined – 2 = Criteria informally defined (embedded in textual descriptions) – 0 = No criteria

Table 1: Existing DQM maturity models in comparison.

Table 1: Existing DQM Maturity Models in Comparison As Table 1 shows, none of the maturity models examined covers all aspects of Enterprise DQM. Guidelines for designing actions for improvement are which represents a collaborative form of DSR offered by two approaches only. Also, all maturity and which is based on having access to and using models examined are characterized by a rigid, practitioners’ knowledge (Österle and Otto 2010). predefined path of development. This, however, stands in contrast with the of DAMA (2009) Fig. 1 gives an overview of view the research approach, that states „[.. ] how each enterprise implements which follows idealized design research processes [DQM] varies widely. Each organisation must (Peffers etanal.implementation 2008; Verschuren and Hartog 2005). determine approach consistent with size, goals, The its research processresources, consists ofand fourcomplexity. activities: However, the essential principles of [DQM] remain analysis, design, evaluation, and diffusion. The the same across the spectrum of enterprises […]”. research is provided by the Competence Taking thiscontext into account, a Maturity Model for Center Corporate Data Quality (CC CDQ) a conEnterprise DQM must provide a dynamic path of development, which each organisation may to sortium research project consisting ofadapt 13 user its individual needs and requirements.

companies, the Institute of Information Management of the University of St. Gallen and the Foundation for Quality Management 4European Research approach (EFQM). Furthermore the research methods draws upon Research method Action Design Research (ADR) as proposed by Sein et al. (2011). ADR addresses interaction The work presented in this paper is the an outcome of design oriented research, following the with practitioners and the organisational context methodological paradigm of Design Science the design artefact is supposed to be used for. Research (DSR). DSR aims at designing artefacts In particular, the maturity model design shares (constructs, models, methods, or instantiations, for example) in orderoftodesign solve and problems occurring in the perception evaluation being practice (Hevner stage et al., 2004;a March Smith, an integrated within designand science re1995). The artefact to be constructed is a maturity search project rather than separated, sequential model that allows to deliberately design Enterprise phases. The integration of building activities, DQM. (organisational) intervention activities, and evaluWhen developing a reliable maturity model a critical ation activities (BIE according to ADR) is depicted factor is the level of maturity of the design domain in Fig. by the bidirectional arrows itself. The2 less developed a design domainconnectis, the higher is the uncertainty in terms of having valid and ing analysis, design, evaluation, and diffusion reliable knowledge about this design domain, and activities. 4.1

the higher is the need for a maturity model that is capable of guiding the path of development for designing the domain. If this is the case, usually 4.2 Research Process only few cases are available that help identify 4.2.1 Analysis possible criteria and evaluate the model, resulting in maturity of limited only. 2006, So access Analysismodels activities began reliability in November comto practitioners‟ knowledge is critical for being able prising the identification of the problem and the to define and evaluate relevant criteria. Therefore, specification of requirements to beselected met by the the overarching research method for designing Maturity ModelEFQM for Enterprise is solution toa be developed. joined theDQM consorconsortium research, represents a tium as a strategic partnerwhich during this first activity collaborative form of DSR and which is based on of the research process, after the decision was having access to and using practitioners‟ knowledge made toand use the 2010). well-established EFQM Model (Österle Otto, Figure 1 gives an overview offorthe research approach, which follows idealized Excellence as a basis for developing the Madesign research processes (Peffers et al., 2008; turity Model for Enterprise DQM. EFQM is a Verschuren and Hartog, 2005). The research process non-profitofaiming establishinganalysis, quality oriented consists four atactivities: design, management Europe. Among other evaluation, and systems diffusion.inThe research context is provided the Competence Center Corporate Data things, by EFQM organizes the annual European Quality (CC CDQ) a consortium research project Quality Award (EQA), in the course of which consisting of 13 user companies, the Institute of companies Management are assessed by of the criteria Information of means the University of St. Gallen and the European Foundation for Quality of the EFQM Model. Relevance of the research to Management (EFQM). be undertaken was confirmed by representatives

from the user of the consortium in Furthermore thecompanies research methods draws upon Action Design Research (ADR) as proposed by Sein a focus group interview and a series of expert et al. (2011). the interaction interviews as ADR well addresses as by a literature analysiswith (cf. practitioners and the organisational context the Related Work). The central outcome of the Anadesign artefact is supposed to be used for. In lysis activity a set ofmodel functional particular, the was maturity designrequirements shares the perception of the design and Model evaluation being an to be met by Maturity as specified by integrated stage within a design science research both the user companies of the consortium and project rather than separated, sequential phases. EFQM. The integration of building activities, (organisational) intervention activities, and evaluation activities (BIE according to ADR) is depicted in Figure 2 by the 4.2.2 Design bidirectional arrows connecting analysis, design, The Maturity Model was built in the course of evaluation, and diffusion activities. three integrated design/evaluate iterations (Fig. 2).

Enterprise Modelling and Information Systems Architectures Vol. X, No. X, Month 200X Enterprise Modelling and Information Systems Architectures A Maturity Model for Enterprise Data Quality Management Vol. 8, No. 2, December 2013 A Maturity Model for Enterprise Data Quality Management

CC CDQ and EFQM agreement

Problem definition by CC CDQ

State of DQM and maturity models

5

9

Requirements of all stakeholder groups

Analysis Scientific publications

Domain

Managerial publications Conferences & seminars

Practical

knowledge

Diffusion

Training material

• Maturity models • DQM practices and

indicators

Scientific knowledge • Maturity Modelling Theory • Maturity Model Design • Capability View on the Firm

Maturity model design

Design

Case studies

Orderly reference modeling

Web-based assessment tool

Evaluation

Focus group interviews

Figure 1: Research process.

4.2

Action research projects

Survey

Figure 1: Research Process

Research process

All three iterations included building activities, 4.2.1 Analysisintervention activities (mainly organisational Analysisaction activities began in November 2006, through research projects), and evaluation comprising the identification of the problem and the activities. The of concrete model design specification requirements to be process met bywas the guided by procedure models for EFQM the development solution to be developed. joined the as a strategic during this ofconsortium maturity models (Becker partner et al. 2010; Bruin et first al. activity of the research process, after the decision 2005). Adaptation mechanisms of reference modwas made to use the well-established EFQM Model elling (Brocke as 2007) allowed systematic for Excellence a basis for developing thedesign Maturity Model for Enterprise DQM. EFQM is a non-profit of the Maturity Model on the basis of the EFQM aiming at establishing quality oriented management Model, following the Guidelines of Modeling systems in Europe. Among other things, EFQM (GoM) (Schuette and Rotthowe 1998). Knowledge organizes the annual European Quality Award (EQA), in thethings coursethat of which arethat assessed about workedcompanies and things did notby means of used the criteria of the Model. work was to draw up EFQM a catalog of Relevance criteria. of the research to be undertaken was confirmed by This knowledge from was gained from related work representatives the user companies of the and from a number of case studies inof consortium in a focus group interviewconducted and a series expert interviews as well as by a literature analysis the context of CC CDQ. (cf. Related work). The central outcome of the Analysis activity was a set of functional requirements to be met by the Maturity Model as 4.2.3 specifiedEvaluation by both the user companies of the consortium and EFQM.

Following the BIE principle of ADR, the evaluation of the Maturity Model was inseparably interwoven with the design of the Model. Evaluation within the three design/evaluate iterations was done by focus groups comprising different

4.2.2

Design

The Maturity Model was built in the course of three stakeholders (organized within consortium workintegrated design/evaluate iterations (Figure 2). All shops) and in the course of tenbuilding action research three iterations included activities, organisational intervention activities through projects (cf. Evaluation). Both ex-ante(mainly and ex-post action research projects), and evaluation activities. evaluation measures were applied, i.e. the artefact The concrete model design process was guided by design theoretical (interior procedure models contribution for the development of mode) maturity models (de Bruin al., 2005; Becker et al., 2010). and its practical useet(external mode) were studied Adaptation mechanisms of reference modelling (vom (Sonnenberg and Vom Brocke 2012). Evaluation Brocke, 2007) allowed systematic design of the activities a survey on the criteria Maturity concluded Model on with the basis of the EFQM Model, following the Guidelines (GoM) (Schuette and maturity levels of of theModeling Maturity Model. A and Rotthowe, 1998). Knowledge about „things that questionnaire was sent to 128 subject matter exworked‟ and „things that did not work‟ was used to perts DQMofdomain, selected drawfrom up athe catalog criteria.who Thiswere knowledge was gained from related work and from a number case from the address database of the Instituteoffor studies conducted in the context of CC CDQ.

Information Management of the University of St. Gallen. of these experts responded, confirm4.2.3 32Evaluation ing the criteria previously identified. of of Following the BIE principle of ADR, the Twenty evaluation the Maturity Model was inseparably interwoven with them declared to be willing to actively support the design of the Model. Evaluation within the three the Maturity Model with their names and the design/evaluate iterations was done by focus groups names of their organisations by means of a joint comprising different stakeholders (organized within consortium with workshops) and in49the course of ten publication EFQM (2011). subject matter action research projects (cf. Evaluation). Both exexperts from 24 user companies, four consulting ante and ex-post evaluation measures were applied, companies, and EFQM joined to evaluate the i.e. the artefact design theoretical contribution (interior mode) and practical useand (external mode) Model. Basically, theits focus groups the survey served to optimize and verify the components and elements of the Maturity Model (in terms

Enterprise Modelling and Information Systems Architectures Vol. X, No. X, Month 200X

Enterprise Modelling and Information Systems Architectures Martin Ofner, Boris Otto, Hubert Österle Vol. 8, No. 2, December 2013 Martin Ofner, Boris Otto, Hubert Österle

6

10

2006

2007

2008

2009

Need articulated in MM evaluated in consortium workshop DD1 AR project DE Iteration 1 Requirements specified in consortium workshop

2010

2011

Cooperation agreed with EFQM MM assessed in consortium workshop

DD2 MM evaluated in AR projects DE Iteration 2 MM evaluated by EFQM DD3 MM evaluated in in AR projects DD4 MM assessed in consortium workshop

MM available for public MM evaluated through survey Web-based assessment tool ready

DE Iteration 3 Legend: MM – Maturity Model; DE – Design/Evaluate; DD – Design Decision.

Figure 2: Design/Evaluate Iterations and Design Decisions.

Figure 2: Design/Evaluate Iterations and Design Decisions

were studied (Sonnenberg and Vom Brocke, 2012). Evaluation activities concluded with a survey on the of optimized wording, whereas the criteria and maturity levelsabove of theall), Maturity Model. A action research projects demonstrating questionnaire was sent aimed to 128at subject matter experts from applicability the DQM domain, who were selected the Model’s and benefit (relating to from the address database of the Institute for the ability to derive improvement actions). Information Management of the University of St. Gallen. 32 of these experts responded, confirming the criteria previously identified. Twenty of them 4.2.4 Diffusion declared to be willing to actively support the Maturity Model with names and the names of The Diffusion phasetheir began in the middle of 2008. their organisations by means of a joint publication The results of the research were disseminated with EFQM (EFQM, 2011). 49 subject matter experts via various publications on from 24 user channels. companies,Scientific four consulting companies, and to evaluate Basically, the EFQM topic joined deal with the gapthe in Model. research to be the focus groups and the survey served to optimize closed, requirements to be met by a maturity and verify the components and elements of the model forModel DQM,(inpossible of application of Maturity terms areas of optimized wording, such aall), model, and the research that above whereas the literature action research projects aimed at demonstrating theetModel‟s applicability was conducted (Hüner al. 2009; Ofner etand al. benefit (relating to the ability to derive improvement 2009).The present paper documents the entire actions).

research process, the design objectives, design 4.2.4 Difussion decisions, and the process of evaluating the arteThe phasebeing began in the middle 2008. fact.Diffusion Apart from documented in of writing, The results of the research were disseminated via the Maturity Model was presented at various various channels. Scientific publications on the topic conferences andgap seminars and was to discussed with deal with the in research be closed, participants, among them the ACM SAC in 2008, requirements to be met by a maturity model for DQM, possible areas of application of such aSystems model, the American Conference of Information (AMCIS) in 2009, the German Information Quality Management Conference (GIQMC) in 2010, and the Stammdaten-Management Forum in 2009 and

and the literature research that was conducted (Ofner et al., 2009; Hüner et al., 2009).The present 2010. Besides, both Maturity Model and the paper documents thethe entire research process, the appraisal methoddesign have decisions, been implemented as a design objectives, and the process of evaluating the artefact. Apart web based assessment tool, which wasfrom made being pubdocumented in writing, the Maturity Model was licly accessible in April 2011 and which allows presented at various conferences and seminars and organisations conduct self-assessments regardwas discussed to with participants, among them the ACM SAC in 2008, American Conference of ing Enterprise DQM.theThe assessment tool also Information Systems (AMCIS) in 2009, theModel. German serves as a platform for diffusion of the Information Quality Management Conference (GIQMC) in 2010, and the StammdatenManagement Forum in 2009 and 2010. Besides, both 5 Model Design the Maturity Model and the appraisal method have 5.1 implemented Scope andasRequirements been a web based assessment tool, which was made publicly accessible in April 2011 The Maturity Modelorganisations for Enterprise aimsselfat and which allows to DQM conduct assessments regarding Enterprisedesign DQMEnter(cf. enabling companies to deliberately https://benchmarking.iwi.unisg.ch). The assessment prise DQM in their organisation. Requirements tool also serves as a platform for diffusion of the to be met bydesign the artefact were identified by the Model. Model

representatives from the user companies of the 2).

4.3 Scope and consortium and byrequirements EFQM (cf. Tab.

The Maturity Model for Enterprise DQM aims at enabling companies to deliberately design Enterprise 5.2 Conceptual Model and Design DQM in their organisation. Requirements to be met Decisions by the artefact were identified by the representatives from the user companies of the Fig. 3 illustrates conceptual of the consortium and bythe EFQM (cf. Table elements 2).

Maturity Model. Model elements adopted from the EFQM Excellence Model are indicated with the EFQM namespace prefix. Tab. 3 lists the design decisions made during different design/evaluate

Enterprise Modelling and Information Systems Architectures

Enterprise Modelling and Information Systems Architectures

Vol. X, No. X, Month 200X Vol. 8, No. 2, December 2013 A Maturity Model for Enterprise Data Quality Management

A Maturity Model for Enterprise Data Quality Management

No.

Requirement

R1

Improvement guidelines: The Maturity Model provides practices how to reach the next, higher level of maturity of Enterprise DQM. It is supposed to be used as a management tool enabling companies to deliberately design Enterprise DQM in their organisation.

R2

Objectivity: The Maturity Model uses a hierarchical model to partition the design domain of Enterprise DQM into smaller entities which can be assessed independently of each other. Also, fuzziness of assessments can be reduced by subdividing the design domain into smaller entities (de Bruin et al., 2005). However, it has to be noted that a maturity model always contains a certain degree of fuzziness.

R3

Dynamic path of development: The Maturity Model is non-prescriptive and allows to identify a dynamic path of development regarding Enterprise DQM. It is important that each company needs to find its own path of development. A maturity model cannot and should not set a predefined path of development to be followed by any company.

R4

Multiple dimensions: The Maturity Model provides multiple dimensions to assess the level of maturity of Enterprise DQM, as progress in organisational change cannot be captured by a single dimension (as progress may refer to the way DQM has been implemented, to business units affected by DQM, etc.).

R5

Assessment methodology: The Maturity Model provides a comprehensive assessment methodology (i.e. a process model, techniques, and tools) for being able to make reliable assessments and to avoid „finger in the wind‟ assessments. The assessment methodology is supposed to allow both selfassessments and assessments by external experts.

R6

Flexibility: The Maturity Model provides configuration mechanisms to reflect specific requirements. It must be applicable for any company, regardless of size or industry. Explicit configuration mechanisms must consistently specify how the Maturity Model may be adapted to company specific requirements.

R7

Conformity with EFQM standard: the EFQM Model for Excellence recognized as an official EFQM ensures connectability with other

The Maturity Model complies with EFQM standards and is based on in order to be adopted into the EFQM model family and to be standard (EFQM, 2003b). Conformity with EFQM standards also methods, techniques, and tools.

Table 2: Functional Requirements to be met by the Model Table 2: Functional requirements to be met by the Model.

iterations, leading to more and model elements being 4.4 Conceptual model design addeddecisions (highlighted with gray background color in Fig.3 3). In the following sections the design Figure illustrates the conceptual elements of the Maturity Model. Model elements decisions are explained in moreadopted detail. Infrom ordertheto EFQM Excellence indicated with the EFQM illustrate every Model designare decision, each explanation namespace prefix. Table 3 lists the design decisions includes a vignette (Stake 1995) giving a concrete made during different design/evaluate iterations, Enterprise oneadded of the leading to DQM more related model example elementsfrom being (highlighted with gray background color in Figure 3). user companies taking part in the action research Inprojects the following sections the design decisions are or in the focus groups. explained in more detail. In order to illustrate every design decision, each explanation includes a vignette 5.2.1 1995) Design decision 1: Use EFQM DQM (Stake, giving a concrete Enterprise related example from one of the Excellence Model as auser basecompanies model taking part in the action research projects or in the focus The groups. first design decision referred to the Maturity Model for Enterprise DQM to be developed on the Design decision 1: Use EFQM Excellence basis ofModel the EFQM Model for Excellence (EFQM as a base model 2009). The EFQM Model is an assessment model The first design decision referred to the Maturity that can be used toDQM identify dynamic path of Model for Enterprise to bea developed on the basis of the EFQM for adopted Excellence (EFQM, development. WhatModel has been in particu2009). The EFQM Model is an assessment model that lar is the overall structure of the EFQM Model can be used to identify a dynamic path of and the content its assessment model, whereas development. Whatofhave been adopted in particular 4.4.1

of of thethe Maturity Model tothe be isthe thedomain overall model structure EFQM Model and content of needs its assessment model, whereasDQM the developed to be filled with Enterprise domain model of the Maturity Model to be developed specific content. Adoption of the EFQM Model’s needs to be „filled‟ with Enterprise DQM specific generic structure ensures compatibility of the content. Adoption of the EFQM Model‟s generic structure of the Maturity Model Maturityensures Modelcompatibility with existing EFQM methods with existing EFQM methods and for and techniques for assessment andtechniques analysis. The assessment and analysis. The assessment assessmentdeveloped dimensions EFQM and dimensions bydeveloped EFQM andbyits partners its partners have been used andreviewed continuously rehave been used and continuously for over twenty years. The content of the domain model of viewed for over twenty years. The content of the the Maturity Model is explicated in the following domain model of the Maturity Model is explicated paragraphs. The Maturity Model is built upon the in thethat following paragraphs. Thedefines Maturity Model logic an organisation that goals for Enterprise DQMthe requires in order is built upon logic certain that ancapabilities organisation that to be able to achieve these goals (cf. Figure 3). At defines goals for Enterprise DQM requires certain its core, the Maturity Model defines 30 Practices and capabilities in Enterprise order to be able to can achieve these 56 Measures for DQM that be used as concrete assessment elements during an appraisal. goals (cf. Fig. 3). At its core, the Maturity Model Whereas Practices are used to assess if and how well defines 30 Practices and 56 Measures for Enterprise certain Enterprise DQM capabilities are established in DQM that can already, be used as concreteallow assessment elean organisation Measures assessing if ments anPractices appraisal. Whereas are and howduring well the support the Practices achievement of Enterprise DQM goals. used to assess if and how well certain Enterprise DQM capabilities are established in an organ-

11

7

Enterprise Modelling and Information Systems Architectures Vol. X, No. X, Month 200X Enterprise Modelling and Information Systems Architectures Martin Ofner, Boris Otto, Hubert Österle Vol. 8, No. 2, December 2013 Martin Ofner, Boris Otto, Hubert Österle

8

12

EFQM domain model Methods and models

posseses

1..1

0..n 1..1 guide creation of 1..n

Capability

1..1 enables achievement of 1..n 0..n

1..1

Design result 1..n

Companyspecific practice grouped by

has 0..n has Goal 1..1

institutionalized in

possible outcome of 1..1

1..n

Organization

measured by

1..n

1..n

EFQM:: Practice

EFQM:: Measure

1..1

Companyspecific measure grouped by

1..1

1..1

1..1

EFQM:: Enabler criterion

grouped by

grouped by

1..n

1..n

1..n EFQM:: Result criterion

EFQM assessment model Companyspecific context category

Assessment context

1..n

1..1 consists 1..n Context value

has 1..n

1..1

Context category

has

1..1

EFQM:: Assessment criterion

EFQM::Maturity level

1..1

1..1 determines

assessed by 1..n EFQM:: Assessment dimension

1..n assigned to 1..1

EFQM:: Score level

1..1

Figure 3: Conceptual model of the Maturity Model.

Figure 3: Conceptual Model of the Maturity Model

Vignette 1. Use EFQM Excellence Model as a base model

isation already, Measures allow assessing if and

A German supplier from the auto industry wants to establish central Enterprise DQM as part of a program for company-wide process harmonization. Certain tasks and activities related to Enterprise DQM are already being done by regional business units. The company now wants to conduct a systematic analysis in order to find out who is doing what already and what needs to be improved. Both the analysis and the continuous improvement process is to be assigned to the company‟s quality management department, which is already using EFQM methods and models.

how well the Practices support the achievement

Another company (from the chemical industry), which established Enterprise DQM as a central management function some years ago, is planning to integrate DQM oriented objectives into the goal structure of certain executive employees. A reliable, standardized methodology is necessary for determining the achievement of objectives to be broadly accepted by the employees affected.

of Enterprise DQM goals.

and future needs to manage enterprise data” is a Practice related to the Enabler criterion 3c (which Vignette 1. Use EFQM Excellence Model as a itself is part of the Enabler criterion 3), or “Success rate of enterprise data quality related training and base model development of individuals” is a Measure related to A German supplier from the automotive inResult criterion “8b. Performance of people results” dustry wants to establish central Enterprise (which itself is part of the Result criterion 8) [for a complete list of and see EFQM DQM as part of Practices a program forMeasures company-wide (2011)]. harmonization. Enabler criteria Certain describetasks whichand areas need process activto be dealt with in order to establish Enterprise ities related to Enterprise DQM are already DQM. “Strategy” addresses leaders to recognize the being done ofbyhigh-quality regional business units. importance enterprise dataThe as a prerequisitenow for wants being able to respond to business company to conduct a systematic drivers (compliance with regulatory and legal analysis in order to find out who is doing what directives, integrated customer management, already what needs to be improved. Both strategic and reporting, or business process integration the and the continuous improvement and analysis standardization, for example). Leaders are required is to of preventive process to bepromote assignedato culture the company’s qualEnterprise DQM. “Controlling” is about the ity management department, which is already quantitative assessment of the quality of enterprise using methods models. data. EFQM Moreover, theand interrelations between Another (from theand chemical industry), enterprisecompany data quality business process performance are Enterprise identified DQM and asmonitored. which established a cent“Organisation & People” ensures that clearly defined ral management function some years ago, is roles, which are specified by clearly defined tasks

planning to integrate DQM oriented objectives into the goal structure of certain executive employees. A reliable, standardized methodology is necessary for determining the achievement of objectives to be broadly accepted by the employees affected.

Enterprise Modelling and Information Systems Architectures

Enterprise Modelling and Information Systems Architectures

Vol. X, No. X, Month 200X Vol. 8, No. 2, December 2013 A Maturity Model for Enterprise Data Quality Management

A Maturity Model for Enterprise Data Quality Management

No.

Design decision

Model elements

DD1

Use EFQM Excellence Model as a base model

EFQM::Enabler criterion, EFQM::Result criterion, EFQM::Practice, EFQM::Measure, EFQM:Assessment dimension, EFQM::Assessment scale, EFQM:Score level, EFQM::Maturity level

DD2

Integrate assessment context

Assessment context, Context category, Category value

DD3

Strengthen common understanding of practices

Design result, Methods and models

DD4

Allow company specific configuration

Company-specific context category, Companyspecific practice, Company-specific measure

Table 3: Overview of Design Decisions (DD) Table 3: Overview of Design Decisions (DD). For reasons of clarity, both Measures and

Practices are hierarchically grouped on two levels and decision-making rights, are assigned to of detail (aspeople. shown in Tab. 4) whereas Measures competent Appropriate assignment of Enterprise DQM responsibilities allows to efficiently are arranged by Result criteria and Practices by and effectively perform DQM related projects and Enabler criteria. To give examples, ‘Running an activities. “Processes and Methods” ensures – adequate DQMDQM training program to through the Enterprise use of Enterprise related processes and services – that knowledge expectationsand arecompetencies fully satisfied develop people’s and that increased value and for future customers and regarding their current needs toother manstakeholders is generated. “Data Architecture” refers age enterprise data’ is a Practice related to the to planning and managing the enterprise data Enabler criterion is part of the architecture in order3c to (which be able itself to ensure enterprise data quality in terms dataofstorage and Enabler criterion 3), of orenterprise ‘Success rate enterprise distribution. “Applications” for Enterprise DQM are data quality related training and development of supposed to provide functionality that supports DQM individuals’ is a Measure related to Result criterion tasks. ‘8b. Performance of people results’ (which itself is Results criteria account for the fact that the way the part of the criterion [for aoncomplete list of of Practices areResult realized has an8)effect the people aPractices company, its customers (including and Measures see EFQM (2011)]). internal Enabler customers, like which e.g. business units project criteria describe areas need to beordealt with teams), the society, and a company‟s overall in order to establish Enterprise DQM. ‘Strategy’ business performance, respectively. EFQM provides addresses recognize the importance an appraisalleaders methodtofor the assessment process, consisting of a procedure model andastechniques for of high-quality enterprise data a prerequisassessment and analysis (EFQM, 2003a). The ite for being able to respond to business drivers appraisal method uses a series of interviews and (compliance regulatory and legal directives, focus groups with as well as document analysis for determining the level of maturity. The most integrated customer management, strategic recomprehensive technique offered is “Results, porting, or business process integration and standApproaches, Deploy, Assess and Refine” (RADAR), ardization, forseven example). Leaders dimensions are requiredfor to which defines Assessment promote and a culture of preventive Enterprise(EFQM, DQM. Practices for Measures, respectively 2009, pp. 22-25). The the level of maturityassessment is always ‘Controlling’ is about quantitative determined according to the same principles, of the quality of enterprise data. Moreover, the regardless of the assessment technique used. For interrelations enterprise dataaquality each Practice between and each Measure score and is determined for eachperformance Assessment dimension usingand an business process are identified Assessment scale. The total result is hierarchically monitored. ‘Organisation and People’ ensures calculated according to predefined calculation that clearly defined are entered specified schemes (EFQM, 2009,roles, p. 27)which and then onby a

clearly defined tasks and decision-making rights, are assigned to competent people. Appropriate 1000-point scale and assigned to one of the three assignment Enterprise Maturity levelsofdefined by theDQM EFQM responsibilities (cf. Figure 4). allows to efficiently and effectively perform DQM 4.4.2 2: Integrate assessment related Design projectsdecision and activities. ‘Processes and context Methods’ ensures through the use of Enterprise As a second design decision it was agreed that the DQMofrelated processes and services—that idea an Assessment context needed expectto be ations are into fullythe satisfied that increased integrated modeland design, as every value single maturity assessment relates to a certain context for customers and other stakeholders is generated. (e.g. management of customer and supplier master ‘Data Architecture’ refers to planning and mandata in regions North America, Europe, and Asia) aging the enterprise dataprior architecture in order that should be predefined to the assessment. What is specified has andata effect on the to be context able to ensure enterprise quality in selection experts todata be interviewed. terms ofofenterprise storage and distribution. ‘Applications’ for Enterprise DQMcontext are supposed Vignette 2. Integrate assessment to provide functionality that supports DQM tasks.

A global provider of telecommunications services aims at criteria establishing Enterprise order be Results account for theDQM factinthat thetoway able to meet the need for high-quality master data the Practices are realized has an effect on the for the new business environment. The company people of a company, management decided its to customers conduct a(including maturity assessment to determine the business current units level or of internal customers, like e.g. maturity of its Enterprise DQM. Toado so, 66 persons project teams), the society, and company’s overfrom six oranisational functions (finance, IT, sales, all business performance, respectively. EFQM etc.) in five countries were selected for being provides anAfter appraisal method for the assessment interviewed. a number of interviews had been conducted the project wondered one process, consisting of agroup procedure modelabout and techinterviewee considering data maintenance processes niques for assessment and analysis (EFQM 2003). to be fully optimized and documented, while another The appraisalsaid method usesprocesses a series ofwere interviews interviewee these badly and focus groups as well asThe document structured and incomplete. reason analysis for this discrepancy was that one interviewee to for determining the level of maturity.referred The most supplier master data for North America, whereas comprehensive technique offered is ‘Results, Apanother interviewee talked about customer master proaches, Assess and Refine’ (RADAR), data for theDeploy, European market. This was taken as an indication that experts always relatedimensions their individual which defines seven Assessment for assessment to a certain context. Practices and for Measures, respectively (EFQM

2009). The level of maturity is always determined

13

9

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 Martin Ofner, Boris Otto, Hubert Österle

14

according to the same principles, regardless of the assessment technique used. For each Practice and each Measure a score is determined for each Assessment dimension using an Assessment scale. The total result is hierarchically calculated according to predefined calculation schemes (EFQM 2009) and then entered on a 1000-point scale and assigned to one of the three Maturity levels defined by the EFQM (cf. Fig. 4).

5.2.2

Design decision 2: Integrate assessment context

As a second design decision it was agreed that the idea of an Assessment context needed to be integrated into the model design, as every single maturity assessment relates to a certain context (e.g. management of customer and supplier master data in regions North America, Europe, and Asia) that should be predefined prior to the assessment. What context is specified has an effect on the selection of experts to be interviewed. If for certain reasons (e.g. limited human resources or budget) certain interview participants cannot be included in the appraisal (e.g. experts for the European and Asian regions are not available), the specified context needs to be revised. It is important that all results recorded from each expert interview or focus group must always be interpreted in relation to the context specified (e.g. when an interviewee’s assessment refers only to customer master data related practices of Enterprise DQM in North America).

In order to be able to consolidate the data collected (from various expert interviews), the context each interview refers to needs to be annotated unambiguously. Three generic context categories plus context values were identified for the Maturity Model: data class, geographic affiliation, and IT system (cf. Fig. 5).

Vignette 2. Integrate assessment context A global provider of telecommunications services aims at establishing Enterprise DQM in order to be able to meet the need for high-quality master data for the new business environment. The company management decided to conduct a maturity assessment to determine the current level of maturity of its Enterprise DQM. To do so, 66 persons from six organisational functions (finance, IT, sales, etc.) in five countries were selected for being interviewed.But one interviewee referred to supplier master data for North America, whereas another interviewee talked about customer master data for the European market. This was taken as an indication that experts always relate their individual assessment to a certain context.

5.2.3

Design decision 3: Strengthen common understanding of practices

The third design decision relates to each Practice being assigned with a set of appropriate Methods and models (plus Design results) allowing to execute each Practice in a structured way. Specifying Design results (strategy documents, measurement systems, etc.) beforehand helps to reduce subjectivity of assessments, as interviewees are given hints as to what type of formal results (documents, templates, reports, systems, etc.) can be expected to result from each Practice. Fig. 5 illustrates the assessment of a Practice and demonstrates how the additional information given about possible Design results strengthens a common understanding. Also, these sets of Methods and models can be used for planning actions for improvement (for a complete list, see (EFQM 2011)).

Enterprise Modelling and Information Systems Architectures Vol. X, No. X, Month 200X A Maturity Model for Enterprise Data Quality Management

11

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 order to be able for to consolidate data collected plus context AInMaturity Model Enterprisethe Data Quality Management

values were identified for the Maturity Model: data class, geographic affiliation, and IT system (cf. Figure 5).

(from various expert interviews), the context each interview refers to needs to be annotated unambiguously. Three generic context categories

EDQM maturity

Measure score 1000

Integrity Causes

Segmentation

Becoming effective

Key results

900 Customer results

800 Comparisons

Performance

People results

Creating structures

700 Trends

Practice score

Overall score

Society results Targets

Soundness

Improvement and innovation Learning and creativity Measurement

600 Applications

500 Data Architecture

400 Processes & Methods

300 Integration 200 Implementation

Organisation & People Controlling

100

System

Design decision 3: Strengthen common understanding of practices

Vignette 3. Strengthen common understandThe third design decision relates to each Practice ing ofassigned Practiceswith a set of appropriate Methods being A leading from the glass industry is and modelscompany (plus Design results) allowing to execute each Practice in a structured way. Specifying Design conducting a maturity assessment of its current results (strategy documents, measurement systems, Enterprise DQM strategy, organisation, and aretc.) beforehand helps reduce subjectivity of chitecture, in order to developare an given actionhints planas forto assessments, as interviewees what type of formal results (documents, templates, improvement. 26 persons from three production reports, systems, etc.) can be expected to result sites in three different countries were selected from each Practice. Figure 5 illustrates the for being a demonstrates group of assessors. assessmentinterviewed of a Practiceby and how the As there was poor common understanding of additional information given about possible Design resultsPractice strengthens a common understanding. each among the assessors, the first Also, asthese sets of Methods and models can be used for sessments conducted were not comparable or planning actions for improvement (for a complete summable. list, cf. (EFQM, 2011)).

Strategy

0

Figure 4: Score assessment andFigure maturity calculation. 4: Score Assessment 4.4.3

Establishing awareness

t1

t2

Time

t3

t4

and Maturity Calculation Vignette 3. Strengthen common understanding of Practices

provided by the Model refer to selection and A leading company from the glass industry is

deselection elements,assessment variation with to conducting aof maturity of itsregard current Enterprise

DQM

strategy,

organisation,

and

architecture, in order toand develop an action plan elefor naming of elements, definition of new improvement. 26 persons from three production

sites in Element three different countries were selected for ments. selection and deselection allows being interviewed by a group of assessors. As there

waslimit poorthe common understanding of each Practice to scope of an assessment by masking among

the

assessors,

the

first

assessments

conducted were notor comparable summable. if the certain Practices Measures.or Especially

Model is used for the first time, it is recommen4.4.4

Design decision 4: Allow company specific

ded to work with a reduced scope. Variation configuration The fourth design decision refers to the Maturity

with to naming of Practices and Measures Modelregard to provide configuration mechanisms, as the

5.2.4

Design decision 4: Allow company specific configuration

The fourth design decision refers to the Maturity Model to provide configuration mechanisms, as the Model is supposed to be applicable to practically any organisation, regardless of size, industry, or individual situation regarding Enterprise DQM. Furthermore, providing configuration mechanisms emphasizes the idea that each organisation should be given the opportunity to find its own path of development with regard to designing Enterprise DQM. Configuration mechanisms

Model is supposed to be applicable to practically any

allows to use regardless synonyms,ofas size, each organisation organisation, industry, or individual

situation

regarding

Enterprise

DQM.

prefers its own, individual terms for denoting Furthermore, providing configuration mechanisms emphasizes the idea that each organisation should

certain in order to to increase the model’s be givenconcepts the opportunity find its own path of development with regard to designing Enterprise

clarity and raise acceptance onprovided the partby of the DQM. Configuration mechanisms the Model refer to selection and deselection of elements,

users. Definition of new elements to filland in variation with regard to naming of allow elements, placeholders in order to add further, individual Company-specific practices, Company-specific measures, or Company-specific context categories.

15

16

company wide level prevents effective Enterprise DQM. As a consequence, the project team decided to establish the Practice “Formalize, review and update Enterprise scope, strategy, objectives, and processes of Enterprise DQM that meets stakeholders‟ needs and expectations and is aligned with the business strategy” together with the Design result “Strategy document” following the Methods and Models of the

P8

PROMET-BSD methodology (IMG, 1998). The “Strategy document” defines the scope, the value contribution, the mandateSystems and a Architectures roadmap for Modelling and Information Enterprise DQM, and is supposed to be verified, Vol. 8,byNo.the 2, December accepted and approved leaders of2013 the company. Martin Ofner, Boris Otto, Hubert Österle

Developing, implementing and improving methods of measurement for enterprise data quality metrics

Assessment context Data class Geographic

Supplier data

Costumer data

Product data

EMEA

NAM

APAC

G1

ERP

HQ

IT System Possible design result

Models and Methods

Measurement system

• Method for specifying business oriented data quality metrics (Hüner, 2011) • Methods and models for performance management (IMG, 1999)

Measurement system to assess data quality and data quality measures by means of metrics. Generally speaking, metrics provide consolidated information on complicated phenomena from the real world on the basis of quantitative measuring. Metric systems are supposed to increase the meaningfulness of individual metrics by structuring them and defining relationships between them.

Assessment Approach

0%

25%

50%

75%

100%

25%

50%

75%

100%

25%

50%

75%

100%

Sound Integrated TOTAL for Approach

37, 5%

Deployment

0%

Implemented

Systematic TOTAL for Deployment

12,5%

Deployment

0%

Measurement Learning and Creativity Improvement and Innovation TOTAL for Deployment

0%

OVERALL TOTAL

25%

Figure 5: Example of a Practice Assessment Form

Figure 5: Example of a practice assessment form.

Vignette 4. Allow company specific configu6 Evaluation ration A German telecommunications provider is planGenerally, evaluating design artefacts must take into ning tothe assess the maturity ofScience its Enterprise account dual nature of Design Research aiming both to advancing DQM at related supplierthe andscientific customerknowledge master base and providing results useful in practice. data maintained by the European ERP system. Sonnenberg and vom Brocke (2012) have identified An glass manufacturer focuses on four international different evaluation types by distinguishing productex-ante masterevaluation data in allinregional and between the course of global artefact design activities and ex-post evaluation during ERP systems with a special interest in pracartefact usage activities. Evaluation type 1 is tices related to data migration projects (due to negative experiences in the past). A German automotive supplier is planning to improve Enterprise DQM maturity in order to reduce the amount of data related process incidents. As these examples show, the Maturity Model is intended to be used by companies from all kinds of industries (chemicals, pharmaceuticals, manufacturing, retail, consumer goods, etc.) and with different experiences made in the past. Each company has its individual assessment context, aims at achieving DQM goals through individual practices, and prefers to use different measures to evaluate whether goals have been

concerned with problem identification, whereas type 6 mainly Demonstration 2 addresses theCase design objectives and the design approach. Evaluation type 3 can be A company, which is one of the world’s leading understood as a proof of the artefact‟s applicability, telecommunications information technology and type 4, finally, as and a proof of its usefulness.

service companies, adapted its business strategy

Evaluation type 1 was mainly addressed by focus in orderand to factor socio-economic groups expert ininterviews during developthe first ments, such as digitalization of project. central The areasneed of design/evaluation iteration of the for maturity modelofwas articulated in late 2006, life,apersonalization products and services, and and specific requirements were revisited in mid increasing mobility of individuals. To validate 2008. Table 5 lists the results of the evaluation of whether the Model. strategy is met on a short-term basis, the Maturity

the company defined a number of goals, such as expanding its leading position in the broadband sector, entering into the entertainment market, or meeting its customers’ expectations with regard to rendering certain products and services. As one measurable objective referring to customer satisfaction it was agreed that customer incidents be reduced by 25% within a year. Business

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 A Maturity Model for Enterprise Data Quality Management

and data management experts of the company supposed that problems in the management of customer data and product data had produced data defects which had a negative impact on business operations, leading to a growing number of customer incidents. Therefore, the company initiated a project to assess the as-is maturity level of Enterprise DQM, identify interrelations between established practices of Enterprise DQM and the impact on the number of customer incidents, and derive improvement actions as deemed appropriate. The project team, which was made up of business and data management experts, selected 30 Practices and three Measures from the Maturity Model for being used in the assessment. Moreover, two Company-Specific practices (e.g. ‘Data integration guidelines are defined, communicated, and applied in relevant projects’) and one Companyspecific measure (‘Number of customer incidents’) were added to take into account the company’s specific requirements, experiences, and goals. The assessment context, which also defines the scope of the assessment, was set to the Context categories ‘Data class’ (‘Customer data’, ‘Product data’), ‘Organisational affiliation’ (‘IT Shared Service department’), and ‘IT System’ (‘Central ERP’) and their respective values. Furthermore, the project team selected ‘RADAR’ as the assessment methodology to be applied (EFQM 2003). Twelve business and IT experts were selected for taking part in interviews in order to determine the assessment scores. The company reached a total score of 305 (out of 1000), calculated as the average of the results for each single criterion (Strategy: 17%; Controlling: 40%; Organisation and People: 27%; Processes and Methods: 42%; Data Architecture: 32%; Applications: 72%; Customer Results: 25%; Society Results: 25%; People Results: 25%; Key Results: 0%; Overall: 30,5%). Hence, at the time of the assessment the company was in the transition process from maturity level one (‘Establishing awareness’) to maturity level two (‘Creating structures’). Both the quantitative results as well as the findings

from the interviews identified strategic deficits as potential root causes of the negative impact of data issues on the Key Results (and the increasing number of customer incidents). For example, it was discovered that the lack of an official mandate (allocated to a company’s department) that allows defining binding rules and guidelines on a company-wide level prevents effective Enterprise DQM. As a consequence, the project team decided to establish the Practice ‘Formalize, review and update scope, strategy, objectives, and processes of Enterprise DQM that meets stakeholders’ needs and expectations and is aligned with the business strategy’ together with the Design result ‘Strategy document’ following the Methods and Models of the PROMET-BSD methodology (IMG 1998). The ‘Strategy document’ defines the scope, the value contribution, the mandate and a roadmap for Enterprise DQM, and is supposed to be verified, accepted and approved by the leaders of the company.

7

Evaluation

Generally, evaluating design artefacts must take into account the dual nature of Design Science Research aiming at both advancing the scientific knowledge base and providing results useful in practice. Sonnenberg and Vom Brocke (2012) have identified four different evaluation types by distinguishing between ex-ante evaluation in the course of artefact design activities and ex-post evaluation during artefact usage activities. Evaluation type 1 is concerned with problem identification, whereas type 2 mainly addresses the design objectives and the design approach. Evaluation type 3 can be understood as a proof of the artefact’s applicability, and type 4, finally, as a proof of its usefulness. Evaluation type 1 was mainly addressed by focus groups and expert interviews during the first design/evaluation iteration of the project. The need for a maturity model was articulated in late 2006, and specific requirements were revisited in mid 2008. Tab. 5 lists the results of the evaluation of the Maturity Model.

17

18

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 Martin Ofner, Boris Otto, Hubert Österle

The design decisions mentioned above were the result of different evaluation types at different stages of the research process. Fig. 2 shows that DD1 (Use of the EFQM Excellence model) resulted from an evaluation of the design approach (type 2) in the course of the first design/evaluation iteration, and that DD2, DD3, and DD4 resulted from evaluation activities taking place in the action research projects (type 3 and 4) in the second and third design/evaluate iteration. Evaluation type 4, i.e. proof of the artefact’s usefulness, was analyzed in greater detail. In particular, the question as to whether the demand for economic efficiency of the Maturity Model is met is difficult to answer. Depending on the scope of the assessment context that was defined, for a project team to apply the Maturity Model in an organisation takes five to thirty days if it is to comprise all phases of the appraisal method (from project preparation to training of staff to deriving actions for improvement) (cf. Tab. 6). Obviously, the effort required for training staff is higher if the Model is used for the first time, and gets lower after repeated use. Applying a maturity model, in general, is a continuous process, for which appropriate organisational structures need to be created. Companies already using EFQM methods and models should be able to quickly understand the Maturity Model and use it regularly, and the staff of companies which have already established quality management should require training with regard to the principles and structures of the EFQM Model only. If there is neither quality management in place nor any knowledge about the EFQM Model at hand, companies need to create adequate organisational structures and build up certain knowledge which may generate substantial costs before they can apply the Maturity Model. From applying the Model some of the companies taking part in the action research projects have derived actions for improvement (ranging from five to twenty), of which some were actually implemented (depending on priorities, budget, or availability of resources).

8 8.1

Conclusions Contribution of the paper

The paper presents a Maturity Model for Enterprise DQM, which aims at supporting enterprises in their effort to deliberately design and establish organisation-wide data quality management. The elements of the Maturity Model are based on principles of quality management in general and existing DQM approaches in particular. The Model’s structure and assessment dimensions have been adopted from the EFQM Model for Excellence. The Model has been approved by EFQM as the official framework for quality oriented management of enterprise data. It comprises, on its most detailed level, 30 practices and 56 measures that can be used as concrete assessment elements during an appraisal. Although the design domain and the purpose of the Maturity Model are specific, findings gained during the artefact design process can be generalized in order to derive further patterns for designing maturity models (e.g. integrating an assessment context). Moreover, through explication of the design process the results can be taken up by other researchers for verification and extension. Furthermore, due to the explication of the design process the model is open to be extended, adapted and reused by future design science research endeavours in related fields. Companies may use the Maturity Model for Enterprise DQM to conduct maturity assessments and derive actions for improvement. Specifying design results to be expected together with taking advantage of appropriate methods and techniques from research and practice is highly useful to support the planning of such actions. The Model’s hierarchical structure allows detailed analysis of the results of a maturity assessment and presentation of these results to different stakeholder groups in an organisation.

8.2

Limitations

The Maturity Model for Enterprise DQM has been used and tested only by large companies so far. Hence, the findings presented in the paper

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 A Maturity Model for Enterprise Data Quality Management

basically apply to the structure and requirements of large companies and cannot be considered to be equally valid for small companies or single company units. Another aspect of limitation refers to the fact that the actions for improvement which were implemented by the companies in the course of action research projects could not be verified (in terms of whether they have actually led to increased DQM maturity). As most of these actions started only recently and are expected to take some time until they start to become effective, the paper does not include any findings on this aspect.

8.3

Need for further research

Further research is expected to refer to continuous maintenance and optimization of the Maturity Model for Enterprise DQM. As the Model is a ‘living’ artefact, it must be reviewed and revised from time to time in order to keep meeting the requirements of different groups (i.e. the scientific community and the practitioners’ community). A web based assessment tool is supposed to facilitate the collection of reference values for levels of maturity regarding Enterprise DQM (best-inclass, industry average, etc.) in order to support the benchmarking process in the future. In this respect, a central challenge lies in finding a balance between the Model’s flexibility and ensuring comparability of results across company boundaries. Furthermore, future research should examine whether the findings presented in the paper can be transferred to other organisational domains and to smaller companies.

References Bach J. (1994) The Immaturity of CMM. In: American Programmer 7(9), pp. 13–18 Baskarada S., Gao J., Koronios A. (2006) Agile Maturity Model Approach to Assessing and Enhancing the Quality of Asset Information in Engineering Asset Management Information Systems. In: Abramowicz W. (ed.) Business Information Systems. Proceedings of Business Infomation Systems, pp. 4–17

Batini C., Scannapieco M. (2006) Data Quality. Concepts, Methodologies and Techniques. Springer, Berlin Batini C., Cappiello C., Francalanci C., Maurino A. (2009) Methodologies for Data Quality Assessment and Improvement. In: ACM Computing Surveys 41(3), 16:1–52 Becker J., Knackstedt R., Pöppelbu¨ss J. (2010) Developing Maturity Models for IT Management? A Procedure Model and its Application. In: Business & Information Systems Engineering 1(3), pp. 213–22 Bitterer A. (2007) Gartner’s Data Quality Maturity Model. Gartner Research Boisot M., Canals A. (2004) Data, information and knowledge: have we got it right? In: Journal of Evolutionary Economics 14(1), pp. 43–67 vom Brocke J. (2007) Design Principles for Reference Modeling: Reusing Information Models by Means of Aggregation, Specialisation, Instantiation, and Analogy In: Reference Modeling for Business Systems Analysis Fettke P., Loos P. (eds.) Idea Group Publishing, Hershey, pp. 47–75 de Bruin T., Freeze R., Kulkarni U., Rosemann M. (2005) Understanding the Main Phases of Developing a Maturity Assessment Model. In: 16th Australasian Conference on Information Systems. Proceedings of the 16th Australasian Conference on Information Systems, pp. 1–11 Christensen C. M., Overdorf M. (2000) Meeting the Challenge of Disruptive Change. In: Harvard Business Review 78(2), pp. 67–75 Crosby P. B. (1979) Quality Is Free - The Art of Making Quality Certain. McGraw-Hill, New York DAMA (2008) The DAMA Dictionary of Data Management. Technics Publications, Denver, Colorado DAMA (2009) The DAMA Guide to the Data Management of Knowledge, First Edition. Technics Publications, Bradley Beach DataFlux (2007) The Data Governance Maturity Model. DataFlux Davenport T. H., Prusak L. (1998) Working

19

20

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 Martin Ofner, Boris Otto, Hubert Österle

Knowledge - How Organizations Manage What They Know. Harvard Business School Press, Boston Dreibelbis A., Hechler E., Milman I., Oberhofer M., van Run P., Wolfson D. (2008) Enterprise Master Data Management: An SOA Approach to Managing Core Information, 1st ed. IBM Press, Upper Saddle River EFQM (2003) The Fundamental Concepts of Excellence. EFQM Press EFQM (2009) EFQM Excellence Model 2010. EFQM Press EFQM (2011) Framework for Corporate Data Quality Management: Assessing the Organizations Data Quality Management Capabilities. EFQM Press, Brussels English L. P. (1999) Improving Data Warehouse and Business Information Quality. John Wiley & Sons, New York Fraser P., Moultrie J., Gregory M. (2002) The use of maturity models / grids as a tool in assessing product development capability. In: IEEE International Engineering Management Conference. Proceedings of the 2002 IEEE International Engineering Management Conference, pp. 244–249 Friedman T. (2006) Gartner Study on Data Quality Shows That IT Still Bears the Burden. In: Gibson C. F., Nolan R. L. (1974) Managing the four stages of EDP growth. In: Harvard Business Review 52(1), pp. 76–88 Gillies A., Howard J. A. (2003) Managing Change in Process and People: Combining a Maturity Model with a Competency-based Approach. In: Total Quality Management & Business Excellence 14(7), pp. 779–787 Hevner A. R., March S. T., Park J., Ram S. (2004) Design Science in Information System Research. In: Management Information Systems Quarterly 28(1), pp. 75–105 Hüner K., Ofner M., Otto B. (2009) Towards a Maturity Model for Corporate Data Quality Management. In: Shin D. (ed.) ACM Symposium on Applied Computing. Proceedings of the 2009 ACM Symposium on Applied Computing, pp. 231–238

IMG (1998) PROMET BSD - Business Strategy Development. Information Management Group, St. Gallen Jones C. (1995) Gaps in SEI Programs. In: Software Development 3(3), pp. 41–48 Kagermann H., Österle H., Jordan J. M. (2010) IT-Driven Business Models. John Wiley & Sons, Hoboken Lee Y. W., Strong D. M., Kahn B. K., Wang R. Y. (2002) AIMQ: a methodology for information quality assessment. In: Information & Management 40, pp. 133–146 Loshin D. (2001) Enterprise knowledge management: the data quality approach. Morgan Kaufmann, San Diego March S. T., Smith G. F. (1995) Design and natural science research on information technology. In: Decision Support Systems 15(4), pp. 251– 266 Mertens P. (2000) Integrierte Informationsverarbeitung (in German: Integrated information processing). Gabler, Wiesbaden Mettler T. (2010) Maturity Assessment Models: A Design Science Research Approach. In: International Journal of Society Systems Science 3(1), pp. 81–98 Nolan R. L. (1973) Managing the Computer Resource: A Stage Hypothesis. In: Communications of the ACM 16(7), pp. 399–405 Ofner M., Hüner K., Otto B. (2009) Dealing with Complexity: A Method to adapt and implement a Maturity Model for Corporate Data Quality Management. In: 15th Americas Conference on Information Systems. Proceedings of the 15th Americas Conference on Information Systems, p. 491 Otto B., Wende K., Schmidt A., Osl P. (2007) Towards a Framework for Corporate Data Quality Management. In: Toleman M., CaterSteel A., Roberts D. (eds.) Proceedings of 18th Australasian Conference on Information Systems. The University of Southern Queensland, pp. 916–926 Paulk M. C., Weber C. W., Garcia S. M., Chrissis M. B., Bush M. (1993) Key Practices of the Capability Maturity Model, Version 1.1. In:

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 A Maturity Model for Enterprise Data Quality Management

Peffers K., Tuunanen T., Rothenberger, A. M., Chatterjee S. (2008) A Design Science Research Methodology for Information Systems Research. In: Journal of Management Information Systems 24(3), pp. 45–77 Pfeffer J., Sutton R. (1999) Knowing "What" To do is not enough: Turning knowledge into action. In: California Management Review 42(1), pp. 83–108 Pierce E., Dismute W. S., Yonke C. L. (2008) The State of Information and Data Governance Understanding How Organizations Govern Their Information and Data Assets. International Association for Information and Data Quality Price R., Shanks G. G. (2005) A semiotic information quality framework: development and comparative analysis. In: Journal of Information Technology (20) 2005, pp. 88–102 Redman T. C. (2000) Data Quality. The Field Guide. Digital Press, Boston Reid A., Catterall M. (2005) Invisible data quality issues in a CRM implementation. In: Journal of Database Marketing & Customer Strategy Management 12(4), pp. 305–314 Ryu K.-S., Park J.-S., Park J.-H. (2006) A Data Quality Management Maturity Model. In: Electronics and Telecommunications Research Institute Journal 28(2) 2006, pp. 191–204 Salchegger M., Dewor E. (2008) Höchste Zeit für ein Solvency II Data Management. In: Versicherungswirtschaft 20, pp. 1714–1718 Saleh Y., Alshawi M. (2005) An Alternative Model for Measuring the Success of IS Projects: The GPIS Model. In: Journal of Enterprise Information Management 18(1), pp. 47– 63 Schuette R., Rotthowe T. (1998) The Guidelines of Modeling - An Approach to Enhance the Quality in Information Models. In: Ling T. W., Lee S. L. (eds.) 17th International Conference on Conceptual Modeling. Proceedings of the 17th International Conference on Conceptual Modeling. Springer, pp. 240–254 SEI (2006a) CMMI for Development, Version 1.2. Carnegie Mellon University, Pittsburgh

SEI (2006b) Standard CMMI Appraisal Method for Process Improvement, Version 1.2. Carnegie Mellon University, Pittsburgh Sein M. K., Henfridsson O., Purao S., Rossi M., Lindgren R. (2011) Action Design Research. In: MIS Quarterly 31(1), pp. 37–56 Shankaranarayan G., Ziad M., Wang R. Y. (2003) Managing Data Quality in Dynamic Decision Environments: An Information Product Approach. In: Journal of Database Management 14(4) 2003, pp. 14–32 Smith H. A., McKeen J. D. (2008) Developments in Practice XXX: Master Data Management: Salvation Or Snake Oil? In: Communications of the Association for Information Systems 23(4), pp. 63–72 Sonnenberg C., Vom Brocke J. (2012) Evaluations in the Science of the Artificial - Reconsidering the Build-Evaluate Pattern in Design Science Research. In: Peffers K. R., A. M., Kuechler B. (eds.) 7th International Conference on Design Science Research in Information Systems and Technology (DESRIST 2012), Las Vegas, NV Spiegler I. (2000) Knowledge management: a new idea or a recycled concept? In: Communications of the AIS 3(4) 374521, pp. 1– 24 Stake R. E. (1995) The Art Of Case Study Research. SAGE Publications, Thousand Oaks Österle H., Otto B. (2010) Consortium Research. A Method for Researcher-Practitioner Collaboration in Design-Oriented IS Research. In: Business & Information Systems Engineering 52(2), pp. 273–285 Tellkamp C., Angerer A., Fleisch E., Corsten D. (2004) From Pallet to Shelf: Improving Data Quality in Retail Supply Chains Using RFID. In: Cutter IT Journal 17(9), pp. 19–24 Vermeer B. H. P. J. (2000) How Important is Data Quality for Evaluating the Impact of EDI on Global Supply Chains? In: 33rd Annual Hawaii International Conference on System Sciences. Proceedings of the 33rd Annual Hawaii International Conference on System Sciences, pp. 1–10 Verschuren P., Hartog R. (2005) Evaluation in

21

Enterprise Modelling and Information Systems Architectures Vol. 8, No. 2, December 2013 Martin Ofner, Boris Otto, Hubert Österle

22

Design-Oriented Research. In: Quality and Quantity 39(6), pp. 733–762 Wand Y., Wang R. Y. (1996) Anchoring Data Quality Dimensions in Ontological Foundations. In: Communications of the ACM 39(11), pp. 86–95 Wang R. Y. (1998) A Product Perspective on Total Data Quality Management. In: Communications of the AIS 41(2), pp. 58–65 Wang R. Y., Strong D. M. (1996) Beyond Accuracy: What Data Quality Means to Data Consumers. In: Journal of Management Information Systems 12(4), pp. 5–34 Wang R. Y., Storey V. C., Firth C. P. (1995) A Framework for Analysis of Data Quality Research. In: IEEE Transactions on Knowledge and Data Engineering 7(4), pp. 623–640 Wang R. Y., Lee Y. W., Pipino L. L., Strong D. M. (1998) Manage Your Information as a Product. In: Sloan Management Review 39(4), pp. 95– 105 Zahay D., Griffin A. (2003) Information antecedents of personalisation and customisation in business-to-business service markets. In: Journal of Database Marketing 10(3), pp. 255– 271 Zhang Z. (2000) Implementation of Total Quality Management. An Empirical Study of Chinese Manufacturing Firms. Dissertation

Martin Ofner Institute of Information Management University of St. Gallen Müller-Friedberg-Str. 8 CH-9000 St. Gallen Switzerland [email protected] Prof. Dr. Boris Otto Professor Audi-Endowed Chair of Supply Net Order Management TU Dortmund University LogistikCampus

Joseph-von-Fraunhofer-Str. 2-4 D-44227 Dortmund Germany [email protected] Prof. Dr. Hubert Österle Professor Institute of Information Management University of St. Gallen Müller-Friedberg-Str. 8 CH-9000 St. Gallen Switzerland [email protected]